RFC: ATA to CAM integration patch

[Alexander Motin]

Mike Tancsa wrote:
> Tried with postmark and bonnie.  Bonnie (no surprise) doesnt show any 
> differences, but postmark shows some
> 
> In AHCI mode
> Time:
>         22 seconds total
>         22 seconds of transactions (681 per second)
> 
> and in stock kernel ATA  mode
> Time:
>         38 seconds total
>         37 seconds of transactions (405 per second)

"Some", you said? :) It's really cool! Have you checked it twice?

> One small bug I noticed is that if I accidentally give camcontrol a 
> bogus value, the disk subsystem seems to lock up
> #camcontrol inquiry 0:0
> 
> On the console, its saying
> ahcich0: Timeout on slot 23
> ahcich0: Timeout on slot 2
> ahcich0: Timeout on slot 14
> ahcich0: Timeout on slot 26
> ahcich0: Timeout on slot 6
> ahcich0: Timeout on slot 18
> ahcich0: Timeout on slot 30

Error recovery is still not finished.

-- 
Alexander Motin